Over 80 percent of clinical trials today fail to meet their stated timelines. Low recruitment/patient enrollment and retention from under-performing sites stalls clinical trials and is costly on a variety of levels. Despite vast and varied efforts, this has not changed greatly over the past two decades. Considering the advances that have been made in many other areas of research with process improvement and technology, e.g. electronic data capture, risk-based monitoring and sophisticated visual analytics, the advances in patient recruitment have not kept pace or yielded the desired benefits.

Those Who Do Not Learn History…

Before considering solutions, one must fully understand the problem, its implications and, for context, have a strong grasp of the current clinical trials environment. Several salient factors can help frame and/or magnify the challenge at hand, most significantly, the indisputable fact that almost 50 percent of clinical trial sites under-perform (defined as not meeting their stated enrollment commitment, in time). According to data from the University of North Carolina at Chapel Hill in 2006: 15-20 percent of sites never enroll a single patient, another 30 percent of sites under-perform, 20 percent of sites are average performers, and 30 percent of sites are high performers.

Fast-forward a decade later and still, according to a recent study published by the Center for the Study of Disease at Duke University, over 10 percent of sites in a given trial will not enroll a single patient. Not surprising as sponsors still largely utilize a number of sophisticated, and some not-so-sophisticated, methods for site selection. Most prominently, reliance on ‘trusted sites’ and often without any metrics or analytics to support that decision. With this approach, subject availability is evaluated via questionnaires to physicians who might become investigators. While questionnaires are necessary, the estimates of available patients are usually inaccurate, so much so, many organizations apply ‘correction factors’ to estimates received to account for well-intentioned, but often misplaced, enthusiasm.

The Proverbial Needle In The Haystack

Compounding the above issue is the fact that medical treatments, and subsequently clinical trials, are becoming more complex. A surrogate for measuring this complexity is the number of eligibility criteria to enroll a subject: the number of major inclusion and exclusion criteria for Phases II and III studies has increased more than 20 percent in recent years. The realization of the long-awaited promise of genomics and proteomics, and the resultant growth of personalized medicines, also has sponsor companies looking for a highly specific patient population to enroll in clinical trials. The net result is a greater number of patients to be screened and a heightened need for targeted recruitment strategies.

To assess all of the eligibility criteria (in some cases, upwards of 50) correctly, investigators must review the medical chart of each possible subject, but there is typically not enough time or resources to effectively complete this. Instead, investigators will hazard a guess or approximation based on past performance. Meanwhile, other organizations, both internal and external to a sponsor, rely on a cadre of tools to identify or recruit patients. These include traditional print and radio/TV advertising, referral networks, social media and other forms of outreach, as well as many methods of data dredging analytics (most of which look to parlay past performance in some predictive fashion). Despite the quantum leaps in technology and data, none have the ability to look at a given site, at a precise moment in time and tell if the exact patient being sought is there. Further compounding this problem is the fact that clinical trial costs per patient continue to rise, being driven by general market conditions such as cost of healthcare delivery and inflation. Additionally, this specific segment is characterized as “ultra competitive,” with more sponsors looking for the same patients, in the same places.

A Costly Gap

There are many process improvement efforts underway to effect speed of study start-up, facilitate more rapid and efficient data collection and improve communications, but still having the right patients with the right diagnosis, signs and symptoms is still at the very core of any successful study. Recent estimates for site start-up costs are between $25,000 to $35,000 (with an additional $1,500 per site, per month for maintenance), whether or not they are productive. When one considers that over three-quarters of clinical trials require extensions to their original timelines, often doubling in length, the cost implications are staggering.

It is easy to point a finger at study design, complexity and the ever increasing number of eligibility criteria but the fact is that sponsors and sites know what these are going into a study. Since these are driven by the clinical and global regulatory landscape, until something changes, a better way to identify the right sites and patients must emerge.

The Future Is Now

The American Medical Informatics Association recently said in comments responding to the FDA Guidance for Industry on EHR use, “More than 96 percent of U.S. hospitals and 83 percent of U.S. office-based physician practices are using EHRs to deliver clinical care, we have an unprecedented opportunity to utilize digitized healthcare data for supplemental uses, such as clinical investigations.” The Patient identification Platform (Patient iP) winner of the 2016 Microsoft Health Innovation Award, was developed to speed trial execution by improving site selection and patient identification.

Designed by an industry leading team of technology and healthcare professionals and guided by a world-class advisory board, Patient iP de-identifies digital health records to locate all appropriate subjects, even against the most restrictive or complex of protocol criteria. With the power of cloud computing and advanced business process logic, this solution has power and speed behind it, evaluating patient records – from tens to hundreds of thousands to millions – for objective criteria matches in real-time. These core abilities allow sponsors to quickly assess a site’s population for eligibility, at a fraction of the time and cost of traditional methods, with much greater precision and accuracy.

The Patient iP platform can evaluate an existing patient population against all active protocols, providing immediate and obvious benefits to the sponsor and sites alike, with output reported as de-identified data at the site level and an aggregate set of master de-identified data at the sponsor (CRO or pharma) level. Unlike the manual processes that this replaces, the evaluation process continues to run in the background – without requiring any further site resources – and provides updated data avoiding recruitment fatigue. New patients, or old patients with changes in healthcare status will be noted on data output reports.

The data output is comprised of two proprietary algorithms – the Patient Matching Index (PMI) which depicts how closely a patient matches trial criteria and the Co-Factor Index (CFI) which depicts how co-existing illness might impact study conduct or outcome and combine to yield a Population Disparity Factor. At the site, the reviewer sees a clear and concise visual presentation of the de-identified data, providing the opportunity for the study staff to review, in a risk-adjusted fashion, an entire population in minutes. Aggregated data displayed in the master de-identified database improves subject selection for trial inclusion, exposes impact of co-morbid illnesses on outcome and monitors each subject over trial progress.

Eliminating Overhead

When employed in advance of study start-up, adoption of advanced clinical trial matching technology reduces non-productive sites, objectively pre-identifies patients, and enables a sponsor to circumvent traditional feasibility methodologies with objective, real-time data for the purpose of developing study timelines and budgets. Patient iP additionally provides the opportunity for effective modeling of protocols, programs and even portfolios against aggregated datasets.

Sponsors and sites today often find themselves using manual processes around trial participation – by digitizing pre-enrollment efforts, sites can focus more on patients and less on administrative tasks, increasing resource utilization and generating revenue for sites. Sponsors in turn can finally improve overall site productivity and performance, consistently deliver projects on time and on budget, and increase the opportunity for recurring business.

About the Author

Dr. Marc Hoffman is Chief Medical officer of Patient identification Platform where he is responsible for providing clinical leadership around Patient iP’s innovative platform, customer programs and related medical affairs activities. He actively surveys and reports on technological and scientific advances while also monitoring industry trends to drive the company’s strategy. Dr. Hoffman also helps lead the Patient iP Advisory Board, interfacing with other clinical experts in regards to their pressing needs — business, strategic and operational.

Download the PDF